中文摘要：

为预测车辆在交叉口群的旅行时间,首先利用城市交通系统呈现的分布、并发特性,采用时延赋色Petri网（TCPN）进行模块化、层次化的建模,建立了包括输入/输出路段车流、交叉口车流、信号控制,以及交叉口群车流的TCPN模型;其次,利用模型的监视器获取仿真过程中车辆的紧迫程度、进出交叉口群的时戳、进出指定库所的时戳、库所容量等状态信息,在此基础上提出了基于紧迫程度和车流密度的车辆平均速率模糊推理算法,实现了对自由旅行路段上车辆平均速率以及车辆旅行时间的预测。试验结果表明：采用基于TCPN交通流模型和模糊推理相结合的仿真预测方法能够合理地对交叉口群中车辆旅行时间进行预测,并且相对于卡尔曼滤波预测方法,提出的方法平均绝对百分误差和均方根误差累计值分别降低了14.33%和22.98%。

英文摘要：

To forecast the travel time of vehicles in intersection group,the distribution and concurrence features of urban traffic system were used to build the modular and hierarchical TCPN model by using timed colored Petri nets firstly.The proposed model considered input/output road section traffic flow,intersection traffic flow,signal control and TCPN model for vehicles intersection group.Then the monitor of the model obtained the vehicle urgent degree,the timestamp of input/output of intersections,the timestamp of input/output of selected library,sink capacity and other state information in the simulation process.According to the obtained information,the fuzzy reasoning algorithm of average speed of vehicles was proposed based on vehicle urgent degree and density of the traffic flow to achieve the prediction for the average speed of vehicles and vehicle travel time in free travel section.The results show that the simulation prediction method based on the combination of timed colored Petri nets traffic flow model and fuzzy reasoning algorithm can predict a reasonable travel time of vehicles inintersections.Compared with Kalman filter prediction method,the proposed method shows that mean-absolute-percentage errors and accumulated root-mean-squared errors of predicted travel time decreased by 14.33% and 22.98%,respectively.